Association among fraction of exhaled nitrous oxide, bronchodilator response and inhaled corticosteroid type

[27.3%]). Most children (n=73 [39.9%]) had moderate persistent asthma. Increased FeNO was associated with percent change in forced expiratory volume in 1 s (FEV 1 ) after bronchodilator adjusted for allergic rhinitis, parental smoking and ICS type (B=0.08 [95% CI 0.04 to 0.12]; P<0.001). Similarly, FeNO was associated with percent change in forced expiratory flow at 25% to 75% of the pulmonary volume (FEF 25-75 ) after bronchodilator adjusted for parental smoking and ICS type (B=0.13 [95% CI 0.01 to 0.24]; P=0.03). FeNO accounted for only 16% and 9% of the variability in FEV 1 and FEF 25-75 , respectively. Mean- adjusted FeNO was lowest in fluticasone users compared with no ICS (mean difference 18.6 parts per billion [ppb] [95% CI 1.0 to 36.2]) and there was no difference in adjusted FeNO level between ciclesonide and no ICS (5.9 ppb [95% CI −9.0 to 20.8]). CONCLUSION: FeNO levels correlated with bronchodilator response in a regional pediatric asthma centre. However, FeNO accounted for only 16% and 9% of the variability in FEV 1 and FEF 25-75 , respectively. Mean adjusted FeNO varied according to ICS type, suggesting a difference in relative efficacy between ICS beyond their dose equivalents.

were seen in the level 2 regional pediatric asthma centre and underwent spirometry testing in addition to FeNO testing were included.This included children diagnosed with asthma on a course of ICS, as well as new referrals to the clinic who may have been prescribed ICS by their referring physician.The present study was approved by the ethics mechanism at Orillia Soldiers' Memorial Hospital, Orillia, Ontario.

Data acquisition
Data were extracted and recorded on a preconstructed standardized form.Collected data included demographic information such as age, sex, weight, height (plotted on the WHO growth curves for children [8]) and self-identified ethnic background.Clinical data included type and dose of ICS, use of a combination product (long-acting betaagonist with ICS), duration of current ICS dose, use of oral steroids in the past three months and use of a leukotriene receptor antagonist.Literature-defined confounding factors were also obtained including presence or absence of eczema, allergic rhinitis, upper respiratory tract infection at time of testing and exposure to parental smoking.The time of day and season of testing were also recorded.

Spirometry and FeNO testing
Both spirometry and FeNO testing were conducted according to American Thoracic Society (ATS) standards (9).FeNO was performed using an asthma inflammation monitor (NIOX MINO, Aerocrine, Sweden), expressed as parts per billion (ppb).Spirometry was performed using a spirometer (Koko, PDS Instrumentation, USA) and was completed after FeNO testing in all cases.Results included both pre-and postbronchodilator values of FEV 1 and FEF 25-75 , along with percentage change.In all cases, the bronchodilator used was salbutamol 400 μg following ATS standards for bronchodilator testing (10).Bronchodilator was delivered by metered-dose inhaler with an aerochamber fitted for appropriate size.The peak expiratory flow (PEF) and FEV 1 to FVC ratio were also recorded.

Asthma severity and dose equivalence
Asthma severity was determined by information from the chart using Global Initiative for Asthma (GINA) criteria ( 1).This information is recorded at every asthma clinic visit by either the pediatrician or the certified asthma educator.ICS dose was stratified according to low, medium and high dose according to Canadian Thoracic Society (CTS) standards (3).Dose equivalents are not available for children <6 years of age; therefore, the equivalent doses for six to 11 years of age were used (Table 1).

Statistical analysis
Data were entered into SPSS version 21.0 (IBM Corporation, USA) and verified for errors.Demographic and clinical data are presented as number and percentage, or mean ± SD, as appropriate.Model coefficients are presented with their 95% CIs, in which the coefficient B represents the expected change in FeNO for each unit change in the predictor variable.
The primary planned analysis was to determine the association between FeNO and postbronchodilation change in FEV 1 by stepwise linear regression while adjusting for known confounders and demographic data.Covariates were retained in the model if they remained significant at P<0.1.A second stepwise linear regression was performed to determine the association between FeNO and postbronchodilation change in FEF 25-75 .In both models, chosen confounders met observed constraints such as having >10 subjects per variable investigated (11).A third analysis, using a generalized linear model, was conducted to evaluate FeNO according to ICS type, while correcting for equivalent dose and other covariates that were found to be statistically significant.Covariates were added to the model if found to have a statistically significant difference in mean FeNO.The Bonferroni correction was used for multiple comparisons.In all final analyses, P<0.05 was considered to be statistically significant.
FeNO concentration varied according to ICS type (Figure 3).Interestingly, this relationship held when the values were corrected for equivalent dose, allergic rhinitis and weight.Mean-adjusted FeNO was lowest in fluticasone users compared with no ICS (mean

DISCUSSION
We conducted a one-year retrospective study investigating FeNO use in a regional pediatric asthma centre in children with known and suspected asthma.FeNO was associated with FEV 1 and FEF 25-75 response to bronchodilator, which are known parameters of asthma control.These findings confirm previous studies evaluating the association of FeNO with lung function in children with asthma (9,(12)(13)(14).However, FeNO only accounted for 16% and 9% of the variability in FEV 1 and FEF 25-75 response, respectively, indicating a high degree of variation when compared with the gold standard.This is likely due to the nonspecific nature of FeNO as a biomarker, which typically reflects inflammation of the airways from any cause (9).The association between FeNO and traditional parameters of asthma control forms the basis of the ATS recommendations for routine use of FeNO in titrating ICS dose in children (9).Currently, CTS guidelines differ from the ATS in this respect based on identical literature sources ( 7).
An interesting finding was that ICS type was an important covariate when investigating the relationship between FeNO and spirometry results.This variable was further explored with a generalized linear model investigating the relationship between FeNO and ICS type.When correcting for equivalent dosing, FeNO levels were lowest in children receiving fluticasone, but there was no difference in FeNO level between ciclesonide and no ICS.There are several possible explanations for this finding.First, ciclesonide is a prodrug, acting locally at the site of delivery and is less bioavailable than fluticasone (15).Therefore, use of fluticasone may lead to higher basal levels of corticosteroid in situ with twice daily dosing compared with ciclesonide.Second, it could be that ciclesonide was used more predominantly in children with more severe asthma and inflammation (ie, confounding by indication); however, this association was not found in our analysis.There was also no difference between FeNO levels in children receiving beclomethasone or budesonide and no ICS; however, this result is not surprising because few children were using these medications.
Ciclesonide has been promoted due to its relatively low side effect profile, and four children with laboratory evidence of adrenal suppression recovered when switched to ciclesonide from fluticasone (16).Whether this finding is due to the relative efficacy of the drug is unknown and data measuring the overall relative potency across all ICS in children with asthma are lacking.Importantly, there are case reports documenting adrenal suppression in children receiving ciclesonide (17).Moreover, the most recent Cochrane Review of ciclesonide versus other ICS therapy in children with chronic asthma cautioned that equipotency may not exist, and one study comparing fluticasone with ciclesonide (2:1 ratio) found that exacerbations were more common in children receiving ciclesonide, although the grade of evidence was low (18).
Equivalent dosing studies are possible with small numbers of participants, and have been conducted in adult populations of steroid-naive patients with asthma (19).The authors used eosinophil count as a marker for inflammation to generate dose-response curves.In their study, both eosinophil count (R 2 =−0.99) and FeNO (R 2 =−0.97) were found to correlate with fluticasone treatment in a dose-dependent manner.A similar study involving children could be conducted; however, sputum is difficult to obtain in this population and more costly to interpret.Ciclesonide therapy has been shown to reduce FeNO within three days to two weeks of initiating 400 μg of therapy in adults with asthma (20,21).A similar decrease was observed after three days in adults receiving a high dose (800 μg/day) compared with 400 μg/day, suggesting that time to onset is equally effective based on the lower dose range (20).However, this has not been reproduced in children using FeNO.
Our study had several limitations.First, the data were retrospective and, although data are meticulously recorded in our clinic, they were collected from the medical chart and our results may be limited by the retrospective nature of the study.Second, although data regarding last oral steroid use and duration of current ICS type were obtained, we did not obtain data regarding cumulative ICS dose, which could have affected our results.Third, we did not apply a measure of adherence, but this should be considered in future prospective studies.Finally, the population sample was mixed, involving suspected asthmatic patients, including both ICS users and ICS-naive patients.This may have limited the power of our analysis, but perhaps adds to the generalizability of our findings.The inclusion of steroid-naive patients enabled a comparison of FeNO levels between ICS type and individuals not on ICS.

CONCLUSION
FeNO levels correlated with known parameters of asthma control in a regional pediatric asthma centre.However, FeNO accounted for only 16% and 9% of the variability in FEV 1 and FEF 25-75 , respectively.Mean-adjusted FeNO varied according to ICS type, suggesting there was a difference in relative efficacy between ICS beyond their dose equivalents.Future studies may be designed with FeNO to explore the relative efficacy of ICS in children with asthma.
ACKNOWLEDGEMENTS: RWS, KD, NS and WGS contributed to study design, implementation and data collection.The statistical analysis was performed by RWS, and supervised by SD and WGS.RWS prepared the first draft of the manuscript, and all authors participated in critical revision before submission for publication.WGS was the supervising author and takes responsibility for the integrity of the data and accuracy of the analysis.

DISCLOSURES:
The authors have no financial disclosures or conflicts of interest to declare.

Figure 1 )Figure 2 )Figure 3 )
Figure 1) Distribution of inhaled corticosteroid (ICS) dose according to type of steroid among children in the pediatric asthma centre (n=43 children not on ICS; † n=2 children on combination product of salmerterol and fluticasone, and were included in the fluticasone group)

Table 1 low, medium and high daily doses of inhaled corticosteroids in micrograms per day* Inhaled corticosteroid Children 6 to 11 years of age adolescents >11 years of age
(1)tained from Global Initiative on Asthma guidelines(1)and included for comparison with older literature difference 18.6 ppb [95% CI 1.0 ppb to 36.2 ppb]; P=0.03) and there was no statistically significant difference in adjusted FeNO concentration between ciclesonide and no ICS (mean difference 5.9 ppb [95% CI −9.0 ppb to 20.8 ppb]; P>0.99).Similarly, the comparison between both beclomethasone and budesonide with no ICS was not statistically significant (mean difference 6.6 ppb [95% CI −10.3 to 23.5 ppb] and 17.6 ppb [95% CI −5.0 to 40.2], respectively).